Internship | Recurrent deep learning applied to radar data

What will be your role? Nowadays, deep learning techniques are used in a wide variety of domains, including the radar domain. In the radar domain, deep learning is primarily applied for classification based on some 2D representation of the radar data, e.g., an Inverse Synthetic Aperture Radar (ISAR) image or a spectrogram (i.e., an image of signal...

What will be your role? Nowadays, deep learning techniques are used in a wide variety of domains, including the radar domain. In the radar domain, deep learning is primarily applied for classification based on some 2D representation of the radar data, e.g., an Inverse Synthetic Aperture Radar (ISAR) image or a spectrogram (i.e., an image of signal...

What will be your role? Deep learning techniques usually thrive when large sets of labelled data are available. In the radar domain, labelled data tend to be scarce depending on the application area. By exploiting expert knowledge it may be possible to improve the training process and overall performance when using only small labelled data sets. From...

What will be your role? Deep learning techniques usually thrive when large sets of labelled data are available. In the radar domain, labelled data tend to be scarce depending on the application area. By exploiting expert knowledge it may be possible to improve the training process and overall performance when using only small labelled data sets. From...

What will be your role? Deep learning techniques typically thrive when large sets of labelled data are available. In the radar domain, labelled data tend to be scarce depending on the application area. So it would be beneficial if all available data could be used. However, in practice measurements may be made with varying radar waveform settings (such...

What will be your role? Deep learning techniques typically thrive when large sets of labelled data are available. In the radar domain, labelled data tend to be scarce depending on the application area. So it would be beneficial if all available data could be used. However, in practice measurements may be made with varying radar waveform settings (such...

What will be your role? The waveform design is one of the critical aspects in radar performance optimisation. Traditionally chirped waveforms are applied in radar, designed for signal-to-noise ratio optimisation when applying a matched filter for detection. This is however only one criterion that can be used for waveform design. Currently there is a...

What will be your role? The waveform design is one of the critical aspects in radar performance optimisation. Traditionally chirped waveforms are applied in radar, designed for signal-to-noise ratio optimisation when applying a matched filter for detection. This is however only one criterion that can be used for waveform design. Currently there is a...

About this position Federated learning (FL) is a hot research topic enabling training on databases of multiple organizations while preserving the privacy of people whose personal data is stored in the databases. FL supports sharing of deep learning (DL) models between the different organizations during the training process without sharing personal...

About this position Federated learning (FL) is a hot research topic enabling training on databases of multiple organizations while preserving the privacy of people whose personal data is stored in the databases. FL supports sharing of deep learning (DL) models between the different organizations during the training process without sharing personal...

About this position Passive radars are defined as a technology that uses signals transmitted by other (communication) systems (i.e., transmitters of opportunity), to perform target detection and parameter estimation (e.g., position, velocity, size, etc.) of the target. Recently, several companies have deployed broadband low-orbit communication...

About this position The Intelligent Imaging department at TNO works on a large variety of computer vision solutions in the field of inspection. One of these projects is the automated inspection of the Dutch highways, where our algorithms are used to monitor the state of the highways and to indicate when sections of highway need to be repaired or...

About this position The Intelligent Imaging department at TNO works on a large variety of computer vision solutions in the field of inspection. One of these projects is the automated inspection of the Dutch highways, where our algorithms are used to monitor the state of the highways and to indicate when sections of highway need to be repaired or...

About this position Image analysis has been revolutionized by the introduction of artificial intelligence and deep learning technology. Robustness of AI-technology is crucial in the high-stakes decision making fields of our defence, safety, and security organisations. There is a rising use of adversarial AI attacks, such as poisoning, which thwarts...

About this position Image analysis has been revolutionized by the introduction of artificial intelligence and deep learning technology. Robustness of AI-technology is crucial in the high-stakes decision making fields of our defence, safety, and security organisations. There is a rising use of adversarial AI attacks, such as poisoning, which thwarts...

About this position Frequency selectivity is crucial for radar system. Spurious frequencies of the higher order harmonics generated in the chain of a front-end radar system can interfere or even damage other systems located in the vicinity. Frequency filtering operations are usually implemented along the RF chain of a transmitter (TX) by electronic...

About this position Frequency selectivity is crucial for radar system. Spurious frequencies of the higher order harmonics generated in the chain of a front-end radar system can interfere or even damage other systems located in the vicinity. Frequency filtering operations are usually implemented along the RF chain of a transmitter (TX) by electronic...

About this position Vision-language models for object detection, with early multimodal fusion mechanisms, are known to have issues regarding false positives. Can we improve the calibration of the object scores, in order to reduce such errors? The paper “Multi-Modal Classifiers for Open-Vocabulary Object Detection” ( improved such models to be...

About this position Vision-language models for object detection, with early multimodal fusion mechanisms, are known to have issues regarding false positives. Can we improve the calibration of the object scores, in order to reduce such errors? The paper “Multi-Modal Classifiers for Open-Vocabulary Object Detection” ( improved such models to be...

About this position Foundation models are AI models that are trained on a large amount of data and can be adapted to a wide range of applications in different domains ( The most wide-known foundation model is currently ChatGPT, developed by OpenAI. Other foundation models available to the public are DALL-E and BERT. Some models are language-only...